Histogram Tissue Characterization of the Fetal Lung as a Noninvasive Indicator of Fetal Lung Maturity
Wael Saied Mohamed Tohamy;
Abstract
Determination of fetal maturity continues to be a crucial factor in obstetric management of the non-routine patients. Reliance on history (last menstrual period, quickening) and physical examination (fundal height) have in large measure been supplanted or at least heavily augmented by laboratory determinations. The prevailing gold standard is currently the estimation of amniotic fluid lecithin/sphingomyelin ratio (L/S ratio) and the presence, both qualitative and quantitative, of phosphatidyl-glycerol (PG) in amniotic fluid. Each separately and both in concert are taken to be indicators of fetal lung maturity, which in turn a major determinant of the neonate’s course.
Amniocentesis is a relatively safe procedure with only a minor degree of associated discomfort. Nevertheless, amniocentesis is an invasive procedure with finite risks and non-universal patient acceptance. The incidence of fetal demise and/or spontaneous abortion secondary to amniocentesis is generally believed to be less than 1%, even with mid-trimester procedures, with reports ranging from 0.2% to 1.4%.
The present study relies on the finding of a noninvasive means of determining fetal lung maturity chiefly by sonographic methods by the use of tissue characterization techniques. So the study of a mathematical measurement of the pixel histogram tissue characterization of the echogenicity of the fetal lung in comparison to pixel histogram tissue characterization echogenicity of the fetal liver in third trimester (34 weeks to 38 weeks) fetuses as non-invasive predictive method of lung maturity is the goal of this thesis.
During the course of routine obstetric sonography for a broad spectrum of indications, scans were performed of the fetal thorax specifically to image the texture of the fetal lung. Longitudinal sonograms were performed to include sections of both the thorax and abdomen, providing representative samples of both lung and liver tissue.
Sonograms were performed with commercially available real-time equipment using both linear and mechanical sector devices with 3.5- MHz variable-internal-focus and 5-MHz fixed-medium-internal-focus transducers, depending on the patient’s body habitus and the position of the fetus with respect to the maternal abdominal wall.
Two hundred pregnant women of gestational age ranging from 34 weeks to 38 weeks (as recorded by their last menstrual period and ultrasound biometry) were included in the study. Patients were recruited from antenatal care clinic, labor ward and elective cesarean sections in Ain Shams University Maternity Hospital.
2D ultrasound scans were performed using the Medison SonoAce X6 with convex probe 3-5 MHz.
Confirmation of gestational age using fetal biometry (HC, BPD, OFD, AC, FL), calculated fetal weights and Histogram tissue characterization will be performed using a constant region of interest of 5mm box for the fetal lung (in the four-chamber view plane) and fetal liver (in trans-abdominal view plane) or a longitudinal (parasagittal or coronal view) where the fetal lung and liver are on the same plane on both sides of the diaphragm.The ultrasound machine automatically plots graphs of the fetal liver and fetal lung will be plotted with automatic calculation of the mean (M).
The results of mean of the pixel histogram of the lung and liver which are stored in an Excel sheet.
Statistical analysis of data obtained in the study was analyzed using software (STATGRAPHICS Centurion XVI.I) after exclusion of the outliers which is out of 3 SD of the mean and a final count of 99 cases are included for the final study.
The sample data of the 99 cases of the study are subjected to the study of:
1- General descriptive statistics of frequency tabulation, histogram and Box-Whisker diagrams.
2- The central tendency of Count, Mean, Standard deviation, Median, Mode, Coeff. of variation, Minimum, Maximum, Range, Standard skewness and kurtosis.
3- Comparison of the mean Pixel Histogram of each organ to itself in the different age groups was performed.
4- Comparison of mean Pixel Histograms between the lung and liver in the different age groups.
5- Correlation tests between the liver Pixel Histogram and the menstrual age.
6- Correlation tests between the lung Pixel Histogram and the menstrual age.
7- Tests of outliers identification were performed to exclude Pixel Histograms outside the three standard deviations of the mean to smoothen the data fittings.
8- Then the correlation tests were redone after the exclusion of the outliers data to select the best fit model between the gestational age and the Pixel Histograms of each organ.
9- Finally the predictive equations of lung maturation in relation to the menstrual age were obtained.
The patient characteristics of the 99 cases included in the present study showed that the mean menstrual age is 36 +/- 1.49 weeks, amniotic fluid is turbid in 89 percent, 65 percentdid not receive corticosteroids to enhance lung maturity, the Apgar’s score of the neonates of 7 or more occurred in 89 percent after 1 minute and 99 percent after 5 minutes and the average fetal weight of the neonates is 2778 +/- 493 gm.
The mean pixel histogram of the lung is 88.76 +/- 23.7 while the pixel histogram of the liver which is the reference organ is 81.1 +/- 20.7 and there is statistical significant difference between the means of the two organs in addition to the change in the behavior of the density curve with progression of age. The pixel histogram of the lung is lower than that of the liver early in age and the pattern is reversed late in pregnancy where the pixel histogram of the lung is higher.
After the outliers testing of data where data outside the +/- 3SD are excluded, the mean pixel histogram of the lung is 90 pixel and the confidence intervals for the mean is 85.75 to 94.6 and the standard deviation is 25.
The used data after outliers testing in the simple regression model to correlate the pixel histogram of the lung and menstrual age showed there is a very high correlation, positive, linear, significant and has resulted in the derivation of an approximation for the prediction of the pixel histogram of the lung by multiplying the menstrual age by the constant 2.45169. The model appears to be adequate at 95.0 % confidence level.
So, the predicted pixel histogram of the lung at the 32 weeks gestation is 78.5 and the 95.0% confidence limit is 74 to 83 and at the 38 weeks gestation the comparable data are 95 and the confidence limits are 90 to 100%.
Further mathematical manipulation to assess if there is a higher R-squared values can be obtained for better prediction. The results of fitting several curvilinear models to the data, the multiplicative model (Y = X^b) yields the highest R-Squared value with 99.8 % which is higher than the linear model (Y = b*X).
The derived equation for the multiplicative model is (Exp(1.26179*ln(MA)). The predicted pixel histogram of the lung at 32 weeks is 79 and the 96 confidence limit is 76 to 82 while at the 38 weeks it is 101 and its confidence limits are 97 to 106.
This is currently selected model is the multiplicative model which yields the highest R-Squared value of 99.8 %
The present work also, showed that the changes in the pixel histogram with progression of age reflects the morphological changes in the lung tissue from the very compact type to the more loose and spongy pulmonary texture due to expansion of the alveolar system of the lung by local fluid secretion and/or filling by inflow of amniotic fluid into the lung alveoli with continuation of pregnancy.
Further studies correlating the pixel histogram of the lung with differentbiochemical amniotic fluid markers for the assessment of lung maturity is needed to confirm the validity of the pixel histogram of the lung as a standalone non-invasive test for prediction of lung maturity prior to termination of pregnancy.
Amniocentesis is a relatively safe procedure with only a minor degree of associated discomfort. Nevertheless, amniocentesis is an invasive procedure with finite risks and non-universal patient acceptance. The incidence of fetal demise and/or spontaneous abortion secondary to amniocentesis is generally believed to be less than 1%, even with mid-trimester procedures, with reports ranging from 0.2% to 1.4%.
The present study relies on the finding of a noninvasive means of determining fetal lung maturity chiefly by sonographic methods by the use of tissue characterization techniques. So the study of a mathematical measurement of the pixel histogram tissue characterization of the echogenicity of the fetal lung in comparison to pixel histogram tissue characterization echogenicity of the fetal liver in third trimester (34 weeks to 38 weeks) fetuses as non-invasive predictive method of lung maturity is the goal of this thesis.
During the course of routine obstetric sonography for a broad spectrum of indications, scans were performed of the fetal thorax specifically to image the texture of the fetal lung. Longitudinal sonograms were performed to include sections of both the thorax and abdomen, providing representative samples of both lung and liver tissue.
Sonograms were performed with commercially available real-time equipment using both linear and mechanical sector devices with 3.5- MHz variable-internal-focus and 5-MHz fixed-medium-internal-focus transducers, depending on the patient’s body habitus and the position of the fetus with respect to the maternal abdominal wall.
Two hundred pregnant women of gestational age ranging from 34 weeks to 38 weeks (as recorded by their last menstrual period and ultrasound biometry) were included in the study. Patients were recruited from antenatal care clinic, labor ward and elective cesarean sections in Ain Shams University Maternity Hospital.
2D ultrasound scans were performed using the Medison SonoAce X6 with convex probe 3-5 MHz.
Confirmation of gestational age using fetal biometry (HC, BPD, OFD, AC, FL), calculated fetal weights and Histogram tissue characterization will be performed using a constant region of interest of 5mm box for the fetal lung (in the four-chamber view plane) and fetal liver (in trans-abdominal view plane) or a longitudinal (parasagittal or coronal view) where the fetal lung and liver are on the same plane on both sides of the diaphragm.The ultrasound machine automatically plots graphs of the fetal liver and fetal lung will be plotted with automatic calculation of the mean (M).
The results of mean of the pixel histogram of the lung and liver which are stored in an Excel sheet.
Statistical analysis of data obtained in the study was analyzed using software (STATGRAPHICS Centurion XVI.I) after exclusion of the outliers which is out of 3 SD of the mean and a final count of 99 cases are included for the final study.
The sample data of the 99 cases of the study are subjected to the study of:
1- General descriptive statistics of frequency tabulation, histogram and Box-Whisker diagrams.
2- The central tendency of Count, Mean, Standard deviation, Median, Mode, Coeff. of variation, Minimum, Maximum, Range, Standard skewness and kurtosis.
3- Comparison of the mean Pixel Histogram of each organ to itself in the different age groups was performed.
4- Comparison of mean Pixel Histograms between the lung and liver in the different age groups.
5- Correlation tests between the liver Pixel Histogram and the menstrual age.
6- Correlation tests between the lung Pixel Histogram and the menstrual age.
7- Tests of outliers identification were performed to exclude Pixel Histograms outside the three standard deviations of the mean to smoothen the data fittings.
8- Then the correlation tests were redone after the exclusion of the outliers data to select the best fit model between the gestational age and the Pixel Histograms of each organ.
9- Finally the predictive equations of lung maturation in relation to the menstrual age were obtained.
The patient characteristics of the 99 cases included in the present study showed that the mean menstrual age is 36 +/- 1.49 weeks, amniotic fluid is turbid in 89 percent, 65 percentdid not receive corticosteroids to enhance lung maturity, the Apgar’s score of the neonates of 7 or more occurred in 89 percent after 1 minute and 99 percent after 5 minutes and the average fetal weight of the neonates is 2778 +/- 493 gm.
The mean pixel histogram of the lung is 88.76 +/- 23.7 while the pixel histogram of the liver which is the reference organ is 81.1 +/- 20.7 and there is statistical significant difference between the means of the two organs in addition to the change in the behavior of the density curve with progression of age. The pixel histogram of the lung is lower than that of the liver early in age and the pattern is reversed late in pregnancy where the pixel histogram of the lung is higher.
After the outliers testing of data where data outside the +/- 3SD are excluded, the mean pixel histogram of the lung is 90 pixel and the confidence intervals for the mean is 85.75 to 94.6 and the standard deviation is 25.
The used data after outliers testing in the simple regression model to correlate the pixel histogram of the lung and menstrual age showed there is a very high correlation, positive, linear, significant and has resulted in the derivation of an approximation for the prediction of the pixel histogram of the lung by multiplying the menstrual age by the constant 2.45169. The model appears to be adequate at 95.0 % confidence level.
So, the predicted pixel histogram of the lung at the 32 weeks gestation is 78.5 and the 95.0% confidence limit is 74 to 83 and at the 38 weeks gestation the comparable data are 95 and the confidence limits are 90 to 100%.
Further mathematical manipulation to assess if there is a higher R-squared values can be obtained for better prediction. The results of fitting several curvilinear models to the data, the multiplicative model (Y = X^b) yields the highest R-Squared value with 99.8 % which is higher than the linear model (Y = b*X).
The derived equation for the multiplicative model is (Exp(1.26179*ln(MA)). The predicted pixel histogram of the lung at 32 weeks is 79 and the 96 confidence limit is 76 to 82 while at the 38 weeks it is 101 and its confidence limits are 97 to 106.
This is currently selected model is the multiplicative model which yields the highest R-Squared value of 99.8 %
The present work also, showed that the changes in the pixel histogram with progression of age reflects the morphological changes in the lung tissue from the very compact type to the more loose and spongy pulmonary texture due to expansion of the alveolar system of the lung by local fluid secretion and/or filling by inflow of amniotic fluid into the lung alveoli with continuation of pregnancy.
Further studies correlating the pixel histogram of the lung with differentbiochemical amniotic fluid markers for the assessment of lung maturity is needed to confirm the validity of the pixel histogram of the lung as a standalone non-invasive test for prediction of lung maturity prior to termination of pregnancy.
Other data
| Title | Histogram Tissue Characterization of the Fetal Lung as a Noninvasive Indicator of Fetal Lung Maturity | Other Titles | التخطيط البياني الصدوي النسيجي لرئه الجنين كوسيله غير اجتياحيه للتنبأ بالنضج الوظيفي للرئه. | Authors | Wael Saied Mohamed Tohamy | Issue Date | 2014 |
Recommend this item
Similar Items from Core Recommender Database
Items in Ain Shams Scholar are protected by copyright, with all rights reserved, unless otherwise indicated.